Merge pull request !25811 from yao_yf/fault_recover_by_mirror_group

Merge pull request !26180 from jjfeing/master

Merge pull request !25614 from TronZhang/parallel_support_in_ascend

Merge pull request !25391 from ougongchang/profiling_stategy

Merge pull request !26070 from chenlei_autodiff/move_file_utils

If SetNodeOutputType functions forcibly splits into multiple functions, the readability decreases, so it blocks lizard scans

Merge pull request !26055 from yanzhenxiang2020/aicpu_random_seed_to_input

Merge pull request !26030 from chenlei_autodiff/decouple_code

Merge pull request !25410 from yuchaojie/ir_fusion3

Merge pull request !26047 from Margaret_wangrui/fallback_log

Add support for no pipeline accugradient

Add delay tag for fusion op

Optimizer the visite order

add mirror for mini step control

Move the group to attributes

Add gradient_shard control for the mini step

Fix code stype

Fix ut description

Add interface

Merge pull request !25765 from TuDouNi/neighborExchangeV2

1. Let 'isa<T>()' calls 'IsSameTypeId()' when T is final,
   and we add 'final' mark to some final classes derived from Base;

2. Generate type id at compile time using constexpr hash function,
   remove TypeIdManager and related code;

3. Add a tool script 'scripts/check_tid.sh' to check tid uniqueness;

4. Improve type search in 'proto_exporter.cc' and 'convert_utils_py.cc';

5. Add some missed MS_DECLARE_PARENT;

6. Ensure calls of 'parent_t::IsFromTypeId()' are all inlined.

Merge pull request !25415 from liubuyu/master

Merge pull request !25395 from limingqi107/new_actor_runtime

Merge pull request !25082 from DeshiChen/1015_set_kernel_attr

it's unreasonable to change the node when generating kernel json.
instead, it should be set in a pass.

most of the operators in original akg_kernel_attrs_process are not longer used,
so we deleted them, leaving only the "Cast" and "MatMul/BatchMatMul".

Merge pull request !21778 from lianliguang/support-dynamic-slice

Merge pull request !25070 from baihuawei/single_mode_bug

Merge pull request !25026 from fangzehua/dynamic_mae_1014

Merge pull request !22731 from DeshiChen/0901_graphkernel_cpu

only Linux system is supported now.

change the default value of `ENABLE_AKG` to off, and controlled by option `-K`.
the `ENABLE_AKG` is auto enabled when `ENABLE_GPU` or `ENABLE_D` is on.
since now, we can use `ENABLE_AKG` to control the compilation of graphkernel
and akg codes.

fix usage description for option "-K", it should be "[-K on|off]".

LLVM is required by akg for cpu kernels, so AKG for cpu is default disabled now.

Merge pull request !23599 from zhouyaqiang0/complex_ops

Merge pull request !23669 from LiangZhibo/inline

Merge pull request !23352 from 张清华/opt_fallback

Merge pull request !23697 from huanghui/clean-warnings-0917